In our OpenStack Cloud setup, we create Volumes from Glance images, extend them to the desired size, and then launch the volumes as instances. We have installed Compute and Block on a same node, and there are multiple such compute-block nodes. The instances run smoothly, but once in a while, the file-system of the VMs become read-only. The error is resolved by rebooting the system. What can we do so that the error does not appear in the first place ?

PS: Our Images are of RHEL 6.8 and Debian 8.6